Federal Crop Insurance and the Disincentives to Adapt to Extreme Heat

The United States produces 40% of the world’s corn and soybeans. Given its dominant market share, any effect on US production has global repercussions. Annual fluctuations in yields crucially depend on the occurrence of extreme heat as measured by temperatures that exceed 29C and 30C, respectively. We examine whether the highly subsidized US crop insurance gives farmers a disincentive to use all possible adaptation strategies against extreme heat as losses are insured. We estimate a panel regression of county-level yields in 1989-2013 that interacts our exogenous weather measures with the fraction of the planted area in a county that is covered by crop insurance. Insurance coverage increased greatly over our sample period. Insured corn and soybean areas have sensitivities that are, respectively, 67% and 43% larger than uninsured areas. We appreciate help in obtaining and interpreting the RMA data from Erik O’Dongohue at the Economic Research Service of the United States Department of Agriculture. All remaining errors are ours. ♣ Columbia University. Email: [email protected]. ♠ Columbia University and NBER. Email: [email protected]. Temperatures in many areas around the world have increased in the past several decades due to an increase in greenhouse gases from anthropogenic sources (Intergovernmental Panel on Climate Change 2013). Continued increases in greenhouse gas emissions will further increase global temperature over the next century. This prediction is robust across multiple climate scenarios. The practical importance of global average temperature on society may appear to be limited in developed countries, but changes in local weather events can translate into changed frequency of extreme events (Brooks 2013). For these reasons, various aspects of extreme events have recently received a lot of attention both in academic and policy analyses, with a special focus on extreme heat (Schär & Jendritzky 2004, Intergovernmental Panel on Climate Change 2012). For example, most of the predicted impacts from climate change in the United States come from the increased frequency of extreme heat (Climate Prospectus 2014). The societal and economic damages associated with climate change in the United States have been studied in the burgeoning climate economics literature covering various sectors: health (Deschênes & Greenstone 2011), productivity (Graff Zivin & Neidell 2014), and agriculture (Schlenker & Roberts 2009). The predicted impacts in all these studies crucially hinge on the sensitivity to fluctuations in extreme heat. Government programs may both enable or potentially distort societal adaptation behavior. Yet, empirical evidence about the effectiveness of governmental programs and whether they aid or interfere with improving the sensitivity of systems to fluctuations in extreme heat is limited. While humans have adapted to extreme heat over time by installing air conditioning, there is no evidence that agriculture has become less sensitive to extreme heat. There is tremendous technological progress for average yields, which have increased threefold over the last six decades. At the same time, the sensitivity to extreme heat has remained relatively constant or even worsened. For example, studies looking at the effect of extreme heat in a panel setting find similar results to a cross-section or studies that link changes in yields to changes in observed climate trends (Burke & Emerick 2013). This paper devises a tractable empirical framework to examine whether the highly subsidized crop insurance program by the United States government makes farmers more sensitive to changes in extreme heat and thereby limits their ability to cope with extreme heat or adapt to it. Insured farmers might not engage in the optimal protection against harmful extreme The whole notion of extreme weather or climate events may cover a broad facet of climate variability under stable or changing climate conditions. Throughout, we adopt IPPC 2012’s definition of “any occurrence of a value of weather or climate variable above (or below) a threshold value near the upper (or lower) ends (tails) of the range of observed values of the variable.”